OneRail today announced OmniSTAR, an AI-powered delivery decisioning platform developed with NVIDIA accelerated computing and acceleration software. The OmniSTAR platform enables enterprise retailers, wholesalers and distributors to evaluate every available delivery option for each order—across owned fleet, courier, parcel and other modes—and select the one that meets service requirements at the lowest cost in real time.

“For most retailers, optimizing last-mile delivery at scale has not been feasible in real time. OmniSTAR changes that,” said Bill Catania, Co-founder and CEO of OneRail. “Until now, same-day delivery at scale was a capability only large retailers like Amazon could afford to build. By combining NVIDIA cuOpt GPU-accelerated decision optimization engine, NVIDIA cuDF for data processing, and NVIDIA accelerated infrastructure with the proprietary data and operational models we’ve built over millions of deliveries, we can now evaluate all available delivery options for each order and make a decision based on cost and service in the moment. That gives any enterprise retailer a practical way to improve margin while offering the kind of fast, flexible delivery consumers expect.”

Built with NVIDIA AI

OmniSTAR builds on a multi-year collaboration between OneRail and NVIDIA through the NVIDIA Inception program. The relationship goes beyond infrastructure to include direct engagement with the NVIDIA cuOpt engineering team, to optimize last-mile delivery and large-scale logistics problems with CuOpt decision optimization engine.

This collaboration enables OneRail to apply advanced optimization within live delivery operations, where decisions must be made in minutes across multiple fulfillment options simultaneously.

“OneRail’s AI-powered delivery platform demonstrates how AI is benefiting businesses and consumers,” said Azita Martin, Vice President and General Manager of Retail and CPG at NVIDIA. “With CuOpt decision optimization engine and NVIDIA AI, OneRail addresses highly complex logistics challenges at scale, enabling faster delivery and lower logistics cost for retailers.”

OmniSTAR was built to solve a problem that has long resisted practical solutions at enterprise scale. Retailers make thousands of delivery decisions each day, each affecting cost, service levels and margin. Most organizations still rely on static rules or manual planning, limiting their ability to evaluate tradeoffs across all available fulfillment options. The challenge is computational. Delivery routing and mode selection involve a large number of possible combinations that cannot be evaluated quickly using conventional systems. As a result, retailers often operate with incomplete information and absorb avoidable costs.

OmniSTAR addresses this by combining NVIDIA cuOpt, an open GPU-accelerated decision optimization engine that accelerates evaluation of complex routing scenarios, with OneRail’s proprietary delivery pricing and performance dataset. NVIDIA AI infrastructure allows OmniSTAR to process sophisticated routing and mode selection scenarios significantly faster than traditional systems, reducing computation time by as much as 10 times—compressing a 20-minute problem down to under two minutes, and a week-long calculation to roughly two days. Because many delivery decisions must be made within tight operational windows, that speed is what makes real-time decisioning possible.

OneRail provides the operating context. Its dataset, built from millions of deliveries across a network of more than 12 million drivers and over 1,000 logistics partners, reflects actual pricing and performance across delivery modes. This allows OmniSTAR to generate data-informed recommendations based on current market conditions rather than assumptions.

Together, these capabilities enable a single system to evaluate all fulfillment options and select the lowest-cost option that meets service requirements.

Deployed and delivering results

OmniSTAR is currently deployed with select enterprise customers. At US Foods, the model identified delivery configurations that were eroding margins, including shipping low-margin items long distances on high-cost equipment. The company used these insights to adjust pricing and restructure delivery patterns.

As market conditions can change rapidly due to fuel costs, weather, shipping and other variables, the ability to solve problems and run multiple scenario analyses more quickly is critical. Powered by NVIDIA acceleration software, OmniSTAR gives retailers a way to continuously optimize delivery decisions and economics as conditions change:

  • It evaluates all available delivery modes for each shipment and selects the most cost-effective option that meets service requirements.

  • It identifies internal rules and constraints that drive unnecessary cost, such as delivery commitments or routing logic that conflict with margin goals.

  • It provides visibility into how delivery decisions affect item-level profitability.

  • It allows teams to adjust delivery strategies as costs and conditions change.

This gives retailers a practical path to offer fast delivery without absorbing the margin tradeoffs that have historically limited same-day delivery at scale.

About OneRail

OneRail is the AI technology platform orchestrating unified commerce for enterprise retailers, wholesalers, and distributors. Powered by its OmniPoint® platform and backed by a 24/7 U.S.-based Exceptions Assist™ team, OneRail connects inventory, transportation, and shopper in a single real-time transaction, giving the world’s leading brands the delivery infrastructure to compete at scale without building it themselves. In March 2026, FedEx selected OneRail as the technology and network partner powering FedEx SameDay® Local. OneRail is the company behind OmniSTAR, the first AI-powered mode-agnostic delivery decisioning platform. OneRail is headquartered in Orlando, Florida, with global operations. To learn more, visit onerail.com.

Media gallery

About The Author
Business Wire